Prevalence of exposure to bovine viral diarrhoea virus (BVDV) - Core

2 downloads 0 Views 1MB Size Report
(Fulton et al., 2004). Herds having at least one weanling ...... Burke, and Kevin Downing, ICBF in sourcing study herds and sup- plying herd data; Dan Collins, ...
Research in Veterinary Science 100 (2015) 21–30

Contents lists available at ScienceDirect

Research in Veterinary Science j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / r v s c

Prevalence of exposure to bovine viral diarrhoea virus (BVDV) and bovine herpesvirus-1 (BoHV-1) in Irish dairy herds R.G. Sayers a,*, N. Byrne a, E. O’Doherty a, S. Arkins b a b

Animal & Bioscience Research Department, Animal & Grassland Research and Innovation Centre, Teagasc, Moorepark, Fermoy, Co. Cork, Ireland Department of Life Sciences, University of Limerick, Limerick, Ireland

A R T I C L E

I N F O

Article history: Received 29 May 2014 Accepted 22 February 2015 Keywords: BVDv BoHV-1 Prevalence Bulk milk ELISA Spot test Seasonality

A B S T R A C T

Bovine viral diarrhoea virus (BVDV) and bovine herpesvirus 1 (BoHV-1) are contagious bovine viral agents. The objectives of this study were to use quarterly bulk milk and ‘spot’ testing of unvaccinated youngstock to establish the national prevalence of exposure to BVDV and/or BoHV-1 in Irish dairy herds. Seasonality of bulk milk ELISA results was also examined. From a geographically representative population of 305 dairy herds, 88% and 80% of herds yielded mean annual positive bulk milk readings for BVDV and BoHV1, respectively. Of these, 61% were vaccinated against BVDV and 12% against BoHV-1. A total of 2171 serum samples from weanlings having a mean age of 291 days yielded 543 (25%) seropositive for BVDV, and 117 (5.4%) seropositive for BoHV-1. A significant seasonal trend in bulk milk antibody ELISA readings and herd status was recorded for BVDV, with more herds categorised as positive in the latter half of the year. © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction Bovine viral diarrhoea (BVD), caused by BVD virus (BVDV), and infectious bovine rhinotracheitis (IBR), caused by bovine herpesvirus 1 (BoHV-1), are highly contagious viral diseases of cattle (Moennig et al., 2005; Muylkens et al., 2007; Nandi et al., 2009). Both exhibit a worldwide distribution (Lindberg et al., 2006; Thiry et al., 2006) and are listed as notifiable diseases by the Office International des Epizooitic1 (OIE). Although OIE-listed diseases, compulsory national control programmes for BVDV and BoHV-1 do not exist in many countries (Ackermann and Engels, 2006; Heffernan et al., 2009). Where regulation does exist, successful BVDV eradication has been achieved through the use of ‘test and cull’ protocols involving removal of persistently infected (PI) individuals (Heffernan et al., 2009; Lindberg et al., 2006; Moennig et al., 2005; Presi et al., 2011; Ridpath, 2012; Ståhl and Alenius, 2012; Valle et al., 2005). In the case of BoHV-1, vaccination with marker/DIVA (Differentiating Infected from VAccinated) vaccines (Mars et al., 2001; Nandi et al., 2009; Nardelli et al., 2008; van Oirschot, 1999) constitutes the primary method of control and eradication in high prevalence regions. In January 2013, a mandatory national eradication programme for BVD, coordinated by the Animal Health Ireland (AHI),

* Corresponding author. Animal & Grassland Research and Innovation Centre, Teagasc, Moorepark, Fermoy, Co. Cork, Ireland. Tel.: +353(0)2542215; fax: +353(0)2542385. E-mail address: [email protected] (R.G. Sayers). 1 www.oie.int.

was introduced in the Republic of Ireland (Graham et al., 2013). As yet, a co-ordinated approach to BoHV-1 control does not exist in Ireland. In order to determine the necessity for, and measure ongoing success of an eradication programme, it is useful to conduct prevalence studies to obtain baseline data (Heffernan et al., 2009; Lindberg et al., 2006; Lindberg and Alenius, 1999; Paisley et al., 2001). National prevalence studies, however, are often prohibitively expensive (Thrushfield, 2005). The advent of bulk milk testing overcomes this issue and reliable antibody detection bulk milk test procedures have been developed for both BVDV and BoHV-1 (Beaudeau et al., 2001; Nylin et al., 2000). Bulk milk analysis for BVDV antibodies, however, does not readily distinguish between vaccinated and unvaccinated herds (Lindberg et al., 2006). This issue has been overcome in the case of BoHV-1 with the advent of BoHV-1 gE-deleted DIVA vaccines. Due to legislative requirements,2 all BoHV-1 vaccines administered in the Republic of Ireland since December 31, 2004 are DIVA vaccines (Simon, 2004). Additionally, bulk milk BVD antibody readings may reflect historical rather than current herd viral status (Brülisauer et al., 2010; Lindberg and Alenius, 1999). To overcome this issue, it is useful to test unvaccinated homeborn youngstock (weanlings) for antibodies against BVDV, i.e. a ‘spot test’ (Houe, 1992, 1994; Mars and Van Maanen, 2005). Positive antibody readings in this population, once maternal antibodies have dissipated, can be indicative of current

2 Diseases of Animals Act 1966; Control on Animal and Poultry Vaccines Order 2002; S.I. 528 of 2002 www.irishstatutebook.ie.

http://dx.doi.org/10.1016/j.rvsc.2015.02.011 0034-5288/© 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/ 4.0/).

22

R.G. Sayers et al./Research in Veterinary Science 100 (2015) 21–30

or recent viral circulation (Houe, 1992, 1994; Lindberg and Alenius, 1999), and as such provide a useful adjunct to bulk milk testing. Although preliminary surveillance studies have indicated high levels of both BVDV and BoHV-1 in the Irish national cattle population (Cowley et al., 2011, 2012; O’Grady et al., 2008; O’Neill et al., 2009), national prevalence data for BVD and BoHV-1 exposure among a geographically representative sample of Irish dairy farms are not available. In addition, evaluation of longitudinal BVD and BoHV-1 bulk milk data over a single lactation in a predominantly springcalving dairy system has not been reported previously. The primary objective of this study, therefore, was to use bulk milk analysis and spot testing of Irish dairy herds to generate national prevalence data for both BVD and BoHV-1, while investigating the usefulness of this diagnostic strategy in an Irish context. 2. Materials and methods 2.1. Sample population and survey The study was licenced by the Irish Department of Health and Children in 2009, meeting all legislative requirements for research involving animals in the Republic of Ireland at the time of the study. A detailed description of the sample population used in this study is outlined in O’Doherty et al. (2013). Briefly, stratified proportional sampling based on herd size and geographical location was used to randomly select and invite 500 herds from the Irish Cattle Breeding Federations (ICBF) database to partake in the study on a nonincentivised basis. Over the 2009 lactation, four bulk milk samples (23 March, 8 June, 31 August and 2 November) were submitted by post in a standardised kit from each participating farm. Each study farm was visited between October 2009 and January 2010 to collect blood samples by coccygeal venepuncture from 20% of the replacement heifer group (weanlings for spot test) on each farm, with a minimum of five weanling heifers sampled on each farm. All heifers were homeborn and not vaccinated against BVDV. Where possible, only weanlings over 270 days of age were sampled, although not achievable in all cases. Accurate weanling age based on calf registration data was downloaded from the ICBF database. 2.2. Sample analysis Commercially available enzyme linked immunosorbent assay (ELISA) kits were used to test bulk milk samples for the presence antibodies against: (i) BVD p80 (NS3) protein, (Institut Pourquier, France); (ii) Ultrapurified IBR lysate (Institut Pourquier, France) in BoHV-1 unvaccinated herds; and (iii) IBRgE, (IDEXX laboratories, USA) in BoHV-1 vaccinated herds. Weanling serum samples were also tested for antibodies against BVD p80, ultrapurified IBR lysate, and IBRgE with serum adapted positive cut-off values applied as outlined by kit manufacturers (Table 1). All analyses were completed by commercial accredited laboratories; BVD p80 and IBR lysate by

National Milk Laboratories Ltd. (UK), and IBR gE by Enfer Diagnostics Ltd. (Ireland). 2.3. Herd classification Calving data from the ICBF were used to determine calvingseason of each herd (spring-calving and non-spring-calving, i.e. spring-autumn [SA] and year-round [YR]) as described by O’Doherty et al. (2013). Vaccination status (vaccinated [V] and unvaccinated [UV]) was determined by questionnaire, with date of vaccination, product used, and class of animal vaccinated (cows, yearlingheifers, weanlings) recorded. In all cases, kit-manufacturer positive cut-off values were applied to ELISA outputs in order to classify herds as ‘positive’ or ‘negative’. Herds were classified as positive or negative at each of the four sampling time points (longitudinal data). Additionally, a mean annual ELISA result for each herd (herd status data) was calculated to provide an overall bulk milk classification for each herd. Herds were also categorised on the basis of combined BVDV and BoHV-1 bulk milk antibody status, i.e. negative for both viral antibodies, positive for BoHV-1 and negative for BVD, negative for BoHV-1 and positive for BVD, and positive for both viral antibodies. Finally, herds were classified with regard to the presence of seropositive unvaccinated weanlings. Two datasets were constructed with weanlings either categorised ‘positive aged ≥180 days of age’ or ‘positive aged ≥270 days of age’ to both assess and minimise potential interference from maternally derived antibodies (MDAs) (Fulton et al., 2004). Herds having at least one weanling serologically positive for either BVDV or BoHV-1 were classified as having ‘evidence of recent viral circulation’ (RVC) (Houe, 1992; Handel et al., 2011). Herds not recording a positive weanling or recording a positive weanling under either 180 or 270 days of age, depending on the dataset, were classified as ‘not having evidence of recent viral circulation’ (NRVC). 2.4. Data analysis Descriptive analysis and graphical representations were completed in Excel (MS Office 2010). Normality of the data was assessed visually using ladder of powers histograms, with normality of residuals assessed using normal probability plots and kernel density estimate plots constructed in Stata (Version 12). True prevalence was calculated using the Rogan–Gladen estimator in the survey toolbox version 1.04 (www.ausvet.com.au (Cameron, 1999)). Pearson’s chi-squared, Fisher’s exact, univariable and multivariable logistic regression, generalised estimating equations (GEE), multinomial logistic regression, Wilcoxon rank sum, and Hosmer–Lemeshow test of goodness of fit analyses were carried out using Stata (Version 12). Seasonal trends in true prevalence for both diseases were tabulated. In addition, box plots of %inhibition, %S/P, and S/N ratio for BVDp80, IBR lysate, and IBR gE, respectively, at each sampling time

Table 1 ELISA kit performance data and positive cut-off values for BVD and BoHV-1assays used in this study. Test

BVD P80 Milk

IBR Lysate Milk

IBR gE Milk

BVD P80 Serum

IBR lysate Serum

IBR gE Serum

Sensitivity Specificity Positive cut-off (Kit) Within-herd prevalence

95.0% 97.7% ≥55 %Inhibitiona ≥30%d

100% 99.6% ≥25 %S/Pb Not available

72.0–88.4% 100% ≤0.8 S/N ratioc 10.0–15.0%e

97.6% 97.3% >60 %Inhibitiona n/a

98.7% 99.9% >55 % S/Pb n/a

100% >99% ≤0.60 S/N ratioc n/a

a b c d e

%Inhibition = [1 − (OD 450 of analysed sample / mean OD 450 of negative control)] × 100. %S/P = (OD 450 of sample − OD 450 of negative control) / (mean OD 450 of positive control − OD 450 of negative control) × 100. S/N ratio = (sample mean − absorbance 650 nm)/negative control mean. Beaudeau et al., 2001. Wellenberg et al, 1998; Kramps et al., 1994.

R.G. Sayers et al./Research in Veterinary Science 100 (2015) 21–30

point were constructed. Two BVD datasets were examined by GEE and logistic regression, i.e. all study herds regardless of BVD vaccination status, and BVD unvaccinated herds only. Longitudinal data were used for the purposes of GEE analysis. To examine seasonal effects on bulk milk analysis, a univariable analysis of bulk milk results (constructed as both categorical [positive vs. negative] and continuous [ELISA readings] variables) and sampling time point was completed (Woodbine et al., 2009). Examination of additional influences on bulk milk longitudinal data (both categorical and continuous) by a number of independent variables was also completed. Independent variables examined included region (high density dairy vs. low density dairy), herd size (31–65 cows vs. 66–99 cows vs. > 99 cows), calving season (springcalving vs. non-spring-calving), type of farming enterprise (dairy livestock only vs. mixed livestock), vaccination status (V vs. UV), and recent viral circulation (RVC ≥180 days or RVC ≥270 days vs. NRVC ≥180 days or NRVC ≥270 days). A total of four datasets were analysed by GEE to account for BVD vaccination status (V, UV) and differing weanling age groups (≥180 and ≥270 days of age). Logistic regression was used to examine associations between recent viral circulation status (RVC ≥180 or ≥270 days vs. NRVC ≥180 or ≥270 days) and vaccination status (V vs. UV), the dependent variables, and region, calving-season, enterprise-type, herd-size, and annual mean bulk milk herd status (independent variables). Four datasets were constructed to account for BVD vaccination status (V, UV) and positive-weanling age (≥180 or ≥270 days). Multinomial logistic regression analysis was completed on combined BVDV and BoHV-1 bulk milk status with region, herd-size, enterprise-type, calving-season and recent viral circulation status as independent variables. For all GEE analyses, herd was included as a repeated measure and an exchangeable correlation used. A binomial distribution was assumed and a logit link function applied for categorical data; a Gaussian distribution and identity link function were used for continuous data. All regression models were constructed by first completing a univariable analysis. Those variables recording p values of ≤0.15 in univariable analyses were included in multivariable models. A manual backwards elimination with a forward step was used to build models with variables recording p values of ≤0.05 maintained. Second level interactions deemed biologically significant were also included. The overall fit of regression models was assessed using the Hosmer–Lemeshow good of fit test following ordinary logistic regression (categorical variables). Normality of residuals was assessed following logistic regression (categorical variables) and linear regression (continuous variables). To examine differences in bulk milk readings between RVC herds and NRVC herds, a Wilcoxon rank sum analysis was completed for each sampling time point. Herds were examined based on vaccination status (V and UV) for both BVDV and BoHV-1, and a third analysis was completed for BVDV where all herds regardless of vaccination status were included. Both viral circulation infection

23

classifications were examined, i.e. herds with positive weanlings greater than 180 or 270 days of age. 3. Results A total of 312 herds were recruited to the study (Fig. 1), yielding a sufficient sample size to achieve a 95% confidence level and precision of 5% for a national dairy herd population of approximately 18,000 herds with an expected national prevalence of 70%. A complete set of four bulk milk samples was not achieved for four farms and vaccination data were not returned by three farmers. Of the herds recruited to the study, 305 herds were therefore suitable for final analysis. Weanling ages were unavailable for eleven herds and these data were excluded from statistical analysis. Study herds have previously been shown to geographically represent the Irish national dairy farm population (O’Doherty et al., 2013). The distribution of study herds across region, herd size, calving season, and type of enterprise is included in Table 2. Approximately 60% of study participants were vaccinated against BVDV using inactivated vaccines, with 12.5% vaccinating against BoHV-1 using DIVA vaccines. A total of 33 study farms administered vaccines for both BVDV and BoHV-1. 3.1. Prevalence of bulk milk positive herds The apparent prevalence (Ap) of bulk milk antibody positive herds for BVDV and BoHV-1 was approximately 88% (80% in unvaccinated herds) and 80% (78% in unvaccinated herds), respectively. True prevalence (Tp) and 95% CI at each sampling time point and across vaccination status is outlined in Table 3. Concurrent exposure to BVDV and BoHV-1 was detected in 72% of herds, with only 10 herds recording bulk milk seronegative status for BVDV and BoHA-1. Seasonal trends in ELISA readings for each disease are included in Fig. 2. 3.2. Seasonal pattern of bulk milk results Univariable GEE analysis highlighted significant seasonal differences in BVDV and BoHV-1 herd status examined as both categorical and continuous variables (Supplementary Table S1). Multivariable analysis of exposure to BVDV and BoHV-1 as continuous variables highlighted a general increase in ELISA readings as the year progressed for both BVD and BoHV-1 (Table 4). When examined as categorical variables, a significant seasonal effect was only observed for BVD bulk milk herd status and a significant interaction between enterprise type and sampling time point was highlighted (Table 4). Herds with a mixed livestock enterprise, in general, were more likely to record a BVD positive bulk milk result in the latter half of the year. This association was apparent regardless of BVD vaccination status. Visual examination of normal probability plots and kernel density plots of residuals did not highlight evidence of non-normality. Goodness of fit analyses for

Table 2 Distribution of study herds across region, herd size, calving-season and enterprise-type. Regiona (density)

Counties represented

Region 1 (Low) 32.5% Region 2 (High) 67.5%

Carlow, Cavan, Clare, Donegal, Dublin, Galway, Kildare, Laois, Leitrim, Longford, Louth, Mayo, Meath, Monaghan, Offaly, Roscommon, Sligo, Westmeath, Wexford, Wicklow Cork, Kerry, Kilkenny, Limerick, Tipperary, Waterford, Limerick

a b c

Herd size (cows)

Calving season

Enterprisec

31–65

66–99

>99

Spring

SA/YRb

Dairy

Mixed

n = 29 9.5%

n = 26 8.5%

n = 44 14.4%

n = 75 24.6%

n = 24 7.9%

n = 52 17%

n = 46 15.1%

n = 52 17.0%

n = 72 23.6%

n = 82 26.9%

n = 190 62.3%

n = 16 5.2%

n = 88 28.9%

n = 118 38.7%

Regions were chosen to correspond with Irish dairy farm distribution (Sayers et al., 2013) and to represent a natural geographical spread. SA/YR represents Spring-Autumn and Year-Round calving seasons. Type of enterprise was not supplied by a single participant.

24

R.G. Sayers et al./Research in Veterinary Science 100 (2015) 21–30

(a)

Region 1

Region 2

Ulster

Connaught (b)

Leinster

Munster Fig. 1. (a) Location of study herds and, (b) representation of the density of animals on dairy farms in Ireland.

categorical data using ordinary logistic regression yielded nonsignificant values ranging from p = 0.286 to p = 0.878. Examination of individual herd results highlighted 15 herds which recorded elevated BVD readings in August and November samplings (Table 5), four of which (herds 67, 263, 273, 275) may be suggestive of introduction of BVD virus to the lactating herd. Herd 263 reported diarrhoea, fever, and milk drop across the lactating herd over the month of July prior to submission of the August bulk milk sample. An additional herd (herd 142) administered vaccine in September which may account for the elevated reading in November. Remaining herds, although having progressed to positive herd status in either August or November, did not record sufficiently elevated readings to be regarded as biologically significant given a positive cut-off of 55% inhibition. 3.3. Youngstock serological status A total of 2171 serum samples from weanlings having a mean age of 291 days (range 109 to 549) were analysed, with 543 testing seropositive for BVDV and 117 testing seropositive for BoHV-1. The

Table 3 True prevalence (Tp) and 95% confidence interval (CI) of exposure to BVDV and BoHV-1 in Irish dairy herds of varying vaccination status at each sampling time point. Sample date

BVD Tp

All herds March June August November Unvaccinated herds only March June August November gE herds* March June August November

n = 305 90.5 90.5 96.2 94.4 n = 113 81.5 83.4 94.9 93.0

IBR 95% CI (%)

Tp

95% CI (%)

86.4,94.7 86.4,94.7 92.8,99.6 90.8,98.1 73.3,89.8 75.4,91.4 89.0,100.0 86.6,99.3

n = 269 80.2 79.6 77.3 79.6 n = 36 100 100 100 100

75.8,84.7 75.1,84.1 72.6,82.0 75.1,84.1 n/a n/a n/a n/a

* Describes herds vaccinated with a BoHV-1 DIVA vaccine and tested using a gE ELISA.

R.G. Sayers et al./Research in Veterinary Science 100 (2015) 21–30

(i)

25

IELISA output (nhibition %)

3.4. Associations between herd demographics and bulk milk status

100 90 80 70 60 50 40 30 20 10 0 Mar June Aug Nov Mar June Aug Nov (Pos) (Pos) (Pos) (Pos) (Neg) (Neg) (Neg) (Neg) Sample date (BVD herd status)

ELISA output (S/P %)

(ii)

450 400 350 300 250 200 150 100 50 0

3.5. Associations between recent viral circulation status, vaccination, and herd demographics

March June August Nov March June Aug Nov (Pos) (Pos) (Pos) (Pos) (Neg) (Neg) (Neg) (Neg) Sample date (BoHV-1 unvaccinated herd status)

(iii)

1.2 ELISA output (S/N ratio)

Regional differences in both BVDV and BoHV-1 herd classification were highlighted by multivariable GEE analysis (Table 4), though this was not consistent across all models. Study herds in the most dairy dense region of Ireland (Region-2) were almost twice as likely to be categorised as BVDV antibody positive over those in Region-1 when all herds, regardless of vaccination status, were included in the model. The reverse was highlighted for BoHV-1, where herds in Region-1 (the least dairy dense part of Ireland) were found to be almost twice as likely as those in Region-2 to be categorised as positive. Herd size was significantly associated with BoHV-1 herd status, with larger herds (>99 cows) approximately four times more likely than smaller herds to be categorised positive. Finally, vaccination was associated with positive herd status for both BVD (OR = 4.29) and BoHV-1 (OR = 31.88), with vaccinating herds more likely to be categorised positive.

1 0.8 0.6

All models examined highlighted a significant association between BVDV bulk milk antibody status and BVDV RVC, herds having evidence of recent BVDV circulation at least three times more likely to be bulk milk positive than those herds recording no seropositive weanlings (Supplementary Table S2 and Table 6). No such association was highlighted in the case of BoHV-1 bulk milk antibody positive herds. A tendency for larger herds to have RVC for either BVDV or BoHV-1 was highlighted, with non-spring-calving herds also more likely to contain BVDV seropositive weanlings (Table 6). Larger herds were more likely to vaccinate for both BVDV and BoHV-1 in this study population. In addition, herds vaccinating for BVDV were more likely to also vaccinate for BoHV-1 and vice versa (Table 7). There were tendencies for herds that were BoHV-1 bulk milk antibody positive to vaccinate for BVDV and for non-springcalving herds to vaccinate for BoHV-1. 3.6. Multinomial logistic regression analysis

0.4 0.2 0 Mar June Aug Nov Mar June Aug Nov (Pos) (Pos) (Pos) (Pos) (Neg) (Neg) (Neg) (Neg) Sample date (BoHV-1 vaccinated herd status)

Fig. 2. Box plots outlining seasonal trendsa in bulk milk ELISA readings across positive (Pos) and negative (Neg) (i) BVD, (ii) BoHV-1 unvaccinated, and (iii) BoHV-1 vaccinated herds in 2009. a Mar: March; June: June; Aug: August; Nov: November.

age profile and BVDV/BoHV-1 serological test status of study weanlings is included in Fig. 3. At least one seropositive weanling over 180 days of age was identified in 119/294 study herds in the case of BVDV, and 24/294 in the case of BoHV-1. If an age limit of 270 days was applied, 96/294 herds recorded a single seropositive BVDV weanling and 18/294 a BoHV-1 seropositive weanling (Fig. 4). A total of 10 herds recorded weanlings ≥180 days old seropositive for both BVDV and BoHV-1, and 8 herds having concurrently seropositive weanlings if an age limit of ≥270 days was applied.

Multinomial logistic regression highlighted that compared to herds bulk milk antibody negative for both BVDV and BoHV-1, larger herds were more likely to be antibody positive for BoHV-1 and negative for BVDV (OR = 3.70, 95% CI = 1.52, 9.04, P = 0.004) and antibody positive for both BVDV and BoHV-1 (OR = 2.67, 95% CI = 1.23, 5.81, P = 0.013). In addition, compared to antibody negative herds, those operating mixed livestock enterprises tended to be over three times more likely than dairy-only herds to present with exposure to one (OR = 4.04, P = 0.071, BoHV-1; OR = 3.35, P = 0.10, BVDV) or both viral pathogens (OR = 4.84, P = 0.024). 3.7. Wilcoxon rank sum analysis A significant difference was highlighted between RVC and NRVC herds in terms of BVDV bulk milk %inhibition readings when all herds were included in the analysis regardless of vaccination status, with z values ranging from −2.718 to −3.864 (Supplementary Table S3). A similar result was generated for BVDV unvaccinated herds alone, with z values ranging from −3.901 to −4.617. No significant difference in %inhibition was highlighted between BVDV vaccinated RVC vs. NRVC herds, however. An analysis of BoHV-1 yielded similar results, with a significant difference in ELISA outputs highlighted for unvaccinated herds (with the sole exception of the March sample), but no significant difference in ELISA readings between vaccinated RVC and NRVC herds (Supplementary Table S3).

26

R.G. Sayers et al./Research in Veterinary Science 100 (2015) 21–30

Table 4 Multivariable GEE analysis of BVDV and BoHV-1 herd classification. Dependent variable Categorical

Independent variable

Odds ratio

BVD status POSITIVE vs. NEGATIVE (All herds included regardless of vaccination status)

Region 2 vs. Region 1 Vaccinated vs. Unvaccinated Mixed August vs. Mixed March Mixed November vs. Mixed March Mixed August vs. Mixed June

2.02 4.29 3.19 2.04 1.96

BVD classification POSITIVE vs. NEGATIVE (UV herds only) IBR classification POSITIVE vs. NEGATIVE (All herds included)

Mixed August vs. Mixed March Mixed November vs. Mixed March

3.46 2.38

Region 1 vs. region 2 Vaccinated vs. Unvaccinated >99cows vs. 31–65 cows >99cows vs. 66–99cows

Continuous BVD ELISA readings

BoHV-1 ELISA readings (excluding vaccinated herds i.e. those tested using gE)

Region 2 vs. Region 1 Vaccinated vs. Unvaccinated August vs. March November vs. March Mixed June vs. Mixed March Mixed August vs. Mixed March Mixed November vs. Mixed March June vs. March August vs. March November vs. March August vs. June November vs. June November vs. August >99cows vs. 31–65 cows >99cows vs. 66–99cows

1.77 31.88 3.66 4.15 Coa 4.58 10.72 4.53 2.73 4.26 3.04 4.30 22.29 14.29 47.82 −8.01 25.52 33.54 68.33 52.08

Confidence interval (95%)

p value

Model (p value)

1.09,3.84 1.09,8.06 1.54,6.58 1.08,3.83 0.96,4.03

0.027